The voice layer

Hermeneutyx (HMX)

Listening, not simply Hearing

The approach

Words tell us what a conversation is about. Nuances in voice and tone give us insights into how the speaker actually feels about the topic. Hermeneutyx (HMX) interprets sentiment, and maps it into the 6-dimensional Affect space of the Semantic Bathymetry™ datamodel.

  • Multiple AI agents interpret voice patterns in parallel. HMX combines frame-level descriptions of pitch, loudness, sound events and the voice itself before any mapping takes place, so that the cadence of speech is preserved.
  • Tone mapped onto k-space. Voice readings are projected onto directional axes — Attraction, Safety, Potency, Trust and Vitality — together with a standalone valence dimension.
  • Sarcasm is identified using a variety of standard- and proprietary flags which calibrate interpretation and guide the subsequent response.

Try the HMX demo

Infographic: HMX maps voice affect into k-space. It fuses pitch, loudness and sound events, maps tone onto the kA to kF axes plus valence, identifies sarcasm, and targets natural dialogue over acted corpora. Diagram of a streamed voice input passing through a chain of analysis stages into a k-vector and onward to SEMBA, with feedback loops returning to the start of the chain.